X-Git-Url: http://git.megacz.com/?a=blobdiff_plain;f=src%2Forg%2Fxwt%2Fjs%2FLexer.java;h=6e541c18af50cd0ecf94851069d68d3a268737ad;hb=32fa4185423afaf8700228f7864d062ded5ca8e8;hp=4752c649561208234fb369a6a00eeffe2009053c;hpb=2a94571c07db6db966f6bfa32583a72287f4db29;p=org.ibex.core.git diff --git a/src/org/xwt/js/Lexer.java b/src/org/xwt/js/Lexer.java index 4752c64..6e541c1 100644 --- a/src/org/xwt/js/Lexer.java +++ b/src/org/xwt/js/Lexer.java @@ -25,13 +25,16 @@ class Lexer implements Tokens { /** for debugging */ public static void main(String[] s) throws Exception { - Lexer l = new Lexer(new InputStreamReader(System.in), "stdin", 0); - int tok = 0; - while((tok = l.getToken()) != -1) System.out.println(codeToString[tok]); + Lexer l = new Lexer(new InputStreamReader(System.in), "stdin", 0); + int tok = 0; + while((tok = l.getToken()) != -1) System.out.println(codeToString[tok]); } /** the token that was just parsed */ protected int op; + + /** the most recently parsed token, regardless of pushbacks */ + protected int mostRecentlyReadToken; /** if the token just parsed was a NUMBER, this is the numeric value */ protected Number number = null; @@ -39,8 +42,11 @@ class Lexer implements Tokens { /** if the token just parsed was a NAME or STRING, this is the string value */ protected String string = null; - /** the line number of the current token */ - protected int line = 0; + /** the line number of the most recently lexed token */ + private int line = 0; + + /** the line number of the most recently parsed token */ + protected int parserLine = 0; /** the column number of the current token */ protected int col = 0; @@ -50,9 +56,10 @@ class Lexer implements Tokens { private SmartReader in; public Lexer(Reader r, String sourceName, int line) throws IOException { - this.sourceName = sourceName; - this.line = line; - in = new SmartReader(r); + this.sourceName = sourceName; + this.line = line; + this.parserLine = line; + in = new SmartReader(r); } @@ -70,99 +77,67 @@ class Lexer implements Tokens { // Token Subtype Handlers ///////////////////////////////////////////////////////// - private int getKeyword(String s) throws IOException { - char c; - switch (s.length()) { - case 2: c=s.charAt(1); - if (c=='f') { if (s.charAt(0)=='i') return IF; } - else if (c=='n') { if (s.charAt(0)=='i') return IN; } - else if (c=='o') { if (s.charAt(0)=='d') return DO; } - break; - case 3: switch (s.charAt(0)) { - case 'a': if (s.charAt(2)=='d' && s.charAt(1)=='n') return AND; break; - case 'f': if (s.charAt(2)=='r' && s.charAt(1)=='o') return FOR; break; - case 'i': if (s.charAt(2)=='t' && s.charAt(1)=='n') return RESERVED; - case 'n': if (s.charAt(2)=='w' && s.charAt(1)=='e') return RESERVED; - case 't': if (s.charAt(2)=='y' && s.charAt(1)=='r') return TRY; break; - case 'v': if (s.charAt(2)=='r' && s.charAt(1)=='a') return VAR; break; - } break; - case 4: switch (s.charAt(0)) { - case 'b': return s.equals("byte") ? RESERVED : -1; - case 'c': c=s.charAt(3); - if (c=='e') { if (s.charAt(2)=='s' && s.charAt(1)=='a') return CASE; } - else if (c=='r') { if (s.charAt(2)=='a' && s.charAt(1)=='h') return RESERVED; } - return -1; - case 'e': c=s.charAt(3); - if (c=='e') { if (s.charAt(2)=='s' && s.charAt(1)=='l') return ELSE; } - else if (c=='m') { if (s.charAt(2)=='u' && s.charAt(1)=='n') return RESERVED; } - return -1; - case 'g': return s.equals("goto") ? RESERVED : -1; - case 'l': return s.equals("long") ? RESERVED : -1; - case 'n': return s.equals("null") ? NULL : -1; - case 't': c=s.charAt(3); - if (c=='e') { if (s.charAt(2)=='u' && s.charAt(1)=='r') return TRUE; } - else if (c=='s') { if (s.charAt(2)=='i' && s.charAt(1)=='h') return THIS; } - return -1; - case 'w': if (s.equals("with")) return RESERVED; else return -1; - case 'v': if (s.equals("void")) return RESERVED; else return -1; - } break; - case 5: switch (s.charAt(2)) { - case 'a': return s.equals("class") ? RESERVED : -1; - case 'e': return s.equals("break") ? BREAK : -1; - case 'i': return s.equals("while") ? WHILE : -1; - case 'l': return s.equals("false") ? FALSE : -1; - case 'n': c=s.charAt(0); - if (s.equals("const")) return RESERVED; - else if (s.equals("final")) return RESERVED; - return -1; - case 'o': c=s.charAt(0); - if (c == 'c') return s.equals("float") ? RESERVED : -1; - else if (c=='s') return s.equals("final") ? RESERVED : -1; - break; - case 'p': return s.equals("super") ? RESERVED : -1; - case 'r': return s.equals("throw") ? THROW : -1; - case 't': return s.equals("catch") ? CATCH : -1; - } break; - case 6: switch (s.charAt(1)) { - case 'a': return s.equals("class") ? RESERVED : -1; - case 'e': c=s.charAt(0); - if (s.equals("delete")) return RESERVED; - else if (c=='r') return s.equals("return") ? RETURN : -1; - break; - case 'h': return s.equals("throws") ? RESERVED : -1; - case 'o': return s.equals("double") ? RESERVED : -1; - case 's': return s.equals("assert") ? ASSERT : -1; - case 'u': return s.equals("public") ? RESERVED : -1; - case 'w': return s.equals("switch") ? SWITCH : -1; - case 'y': return s.equals("typeof") ? TYPEOF : -1; - } break; - case 7: switch (s.charAt(1)) { - case 'a': return s.equals("package") ? RESERVED : -1; - case 'e': return s.equals("default") ? DEFAULT : -1; - case 'i': return s.equals("finally") ? FINALLY : -1; - case 'o': return s.equals("boolean") ? RESERVED : -1; - case 'r': return s.equals("private") ? RESERVED : -1; - case 'x': return s.equals("extends") ? RESERVED : -1; - } break; - case 8: switch (s.charAt(0)) { - case 'a': return s.equals("abstract") ? RESERVED : -1; - case 'c': return s.equals("continue") ? CONTINUE : -1; - case 'd': return s.equals("debugger") ? RESERVED : -1; - case 'f': return s.equals("function") ? FUNCTION : -1; - case 'v': return s.equals("volatile") ? RESERVED : -1; - } break; - case 9: c=s.charAt(0); - if (c=='i') return s.equals("interface") ? RESERVED : -1; - else if (c=='p') return s.equals("protected") ? RESERVED : -1; - else if (c=='t') return s.equals("transient") ? RESERVED : -1; - break; - case 10: c=s.charAt(1); - if (c=='m') return s.equals("implements") ? RESERVED : -1; - else if (c=='n' && s.equals("instanceof")) return RESERVED; - break; - case 12: return s.equals("synchronized") ? RESERVED : -1; - } - return -1; + private int getKeyword(String name) throws IOException { + //#switch(name) + case "if": return IF; + case "lt": return LT; + case "gt": return GT; + case "in": return IN; + case "do": return DO; + case "and": return AND; + case "for": return FOR; + case "int": return RESERVED; + case "new": return RESERVED; + case "try": return TRY; + case "var": return VAR; + case "byte": return RESERVED; + case "case": return CASE; + case "char": return RESERVED; + case "else": return ELSE; + case "enum": return RESERVED; + case "goto": return RESERVED; + case "long": return RESERVED; + case "null": return NULL; + case "true": return TRUE; + case "with": return RESERVED; + case "void": return RESERVED; + case "class": return RESERVED; + case "break": return BREAK; + case "while": return WHILE; + case "false": return FALSE; + case "const": return RESERVED; + case "final": return RESERVED; + case "super": return RESERVED; + case "throw": return THROW; + case "catch": return CATCH; + case "class": return RESERVED; + case "delete": return RESERVED; + case "return": return RETURN; + case "throws": return RESERVED; + case "double": return RESERVED; + case "assert": return ASSERT; + case "public": return RESERVED; + case "switch": return SWITCH; + case "typeof": return TYPEOF; + case "package": return RESERVED; + case "default": return DEFAULT; + case "finally": return FINALLY; + case "boolean": return RESERVED; + case "private": return RESERVED; + case "extends": return RESERVED; + case "abstract": return RESERVED; + case "continue": return CONTINUE; + case "debugger": return RESERVED; + case "function": return FUNCTION; + case "volatile": return RESERVED; + case "interface": return RESERVED; + case "protected": return RESERVED; + case "transient": return RESERVED; + case "implements": return RESERVED; + case "instanceof": return RESERVED; + case "synchronized": return RESERVED; + //#end + return -1; } private int getIdentifier(int c) throws IOException { @@ -171,7 +146,7 @@ class Lexer implements Tokens { in.unread(); String str = in.getString(); int result = getKeyword(str); - if (result == RESERVED) throw new LexerException("The reserved word \"" + str + "\" is not permitted in XWT scripts"); + if (result == RESERVED) throw new LexerException("The reserved word \"" + str + "\" is not permitted in XWT scripts"); if (result != -1) return result; this.string = str.intern(); return NAME; @@ -211,18 +186,15 @@ class Lexer implements Tokens { if (isInteger) { longval = Long.parseLong(numString, base); dval = (double)longval; - } else { + } else { dval = Double.parseDouble(numString); longval = (long) dval; if (longval == dval) isInteger = true; } } - if (!isInteger) this.number = new Double(dval); - else if (Byte.MIN_VALUE <= longval && longval <= Byte.MAX_VALUE) this.number = new Byte((byte)longval); - else if (Short.MIN_VALUE <= longval && longval <= Short.MAX_VALUE) this.number = new Short((short)longval); - else if (Integer.MIN_VALUE <= longval && longval <= Integer.MAX_VALUE) this.number = new Integer((int)longval); - else this.number = new Double(longval); + if (!isInteger) this.number = JS.N(dval); + else this.number = JS.N(longval); return NUMBER; } @@ -248,20 +220,20 @@ class Lexer implements Tokens { case 't': c = '\t'; break; case 'v': c = '\u000B'; break; case '\\': c = '\\'; break; - case 'u': { - int v = 0; - for(int i=0; i<4; i++) { - int ci = in.read(); - if (!((ci >= '0' && ci <= '9') || (ci >= 'a' && ci <= 'f') || (ci >= 'A' && ci <= 'F'))) - throw new LexerException("illegal character '" + ((char)c) + "' in \\u unicode escape sequence"); - v = (v << 8) | Integer.parseInt(ci + "", 16); - } - c = (char)v; - break; - } + case 'u': { + int v = 0; + for(int i=0; i<4; i++) { + int ci = in.read(); + if (!((ci >= '0' && ci <= '9') || (ci >= 'a' && ci <= 'f') || (ci >= 'A' && ci <= 'F'))) + throw new LexerException("illegal character '" + ((char)c) + "' in \\u unicode escape sequence"); + v = (v << 8) | Integer.parseInt(ci + "", 16); + } + c = (char)v; + break; + } default: - // just use the character that was escaped - break; + // just use the character that was escaped + break; } } if (stringBuf != null) stringBuf.append((char) c); @@ -276,7 +248,7 @@ class Lexer implements Tokens { return STRING; } - public int _getToken() throws IOException { + private int _getToken() throws IOException { int c; do { c = in.read(); } while (c == '\u0020' || c == '\u0009' || c == '\u000C' || c == '\u000B' || c == '\n' ); if (c == -1) return -1; @@ -307,7 +279,7 @@ class Lexer implements Tokens { case '*': return in.match('=') ? ASSIGN_MUL : MUL; case '<': return !in.match('<') ? (in.match('=') ? LE : LT) : in.match('=') ? ASSIGN_LSH : LSH; case '>': return !in.match('>') ? (in.match('=') ? GE : GT) : - in.match('>') ? (in.match('=') ? ASSIGN_URSH : URSH) : (in.match('=') ? ASSIGN_RSH : RSH); + in.match('>') ? (in.match('=') ? ASSIGN_URSH : URSH) : (in.match('=') ? ASSIGN_RSH : RSH); case '/': if (in.match('=')) return ASSIGN_DIV; if (in.match('/')) { while ((c = in.read()) != -1 && c != '\n'); in.unread(); return getToken(); } @@ -334,9 +306,11 @@ class Lexer implements Tokens { public SmartReader(Reader r) { reader = new PushbackReader(r); } public void unread() throws IOException { unread((char)lastread); } public void unread(char c) throws IOException { - reader.unread(c); - if (accumulator != null) accumulator.setLength(accumulator.length() - 1); - } + reader.unread(c); + if(c == '\n') col = -1; + else col--; + if (accumulator != null) accumulator.setLength(accumulator.length() - 1); + } public boolean match(char c) throws IOException { if (peek() == c) { reader.read(); return true; } else return false; } public int peek() throws IOException { int peeked = reader.read(); @@ -346,19 +320,23 @@ class Lexer implements Tokens { public int read() throws IOException { lastread = reader.read(); if (accumulator != null) accumulator.append((char)lastread); - if (lastread != '\n' && lastread != '\r') col++; - if (lastread == '\n') { line++; col = 0; } + if (lastread != '\n' && lastread != '\r') col++; + if (lastread == '\n') { + // col is -1 if we just unread a newline, this is sort of ugly + if (col != -1) parserLine = ++line; + col = 0; + } return lastread; } // FEATURE: could be much more efficient StringBuffer accumulator = null; public void startString() { - accumulator = new StringBuffer(); - accumulator.append((char)lastread); - } + accumulator = new StringBuffer(); + accumulator.append((char)lastread); + } public String getString() throws IOException { - String ret = accumulator.toString(); + String ret = accumulator.toString().intern(); accumulator = null; return ret; } @@ -372,45 +350,48 @@ class Lexer implements Tokens { private Object[] pushBackObjects = new Object[10]; /** push back a token */ - public void pushBackToken(int op, Object obj) { - if (pushBackDepth >= pushBackInts.length - 1) { - int[] newInts = new int[pushBackInts.length * 2]; - System.arraycopy(pushBackInts, 0, newInts, 0, pushBackInts.length); - pushBackInts = newInts; - Object[] newObjects = new Object[pushBackObjects.length * 2]; - System.arraycopy(pushBackObjects, 0, newObjects, 0, pushBackObjects.length); - pushBackObjects = newObjects; - } - pushBackInts[pushBackDepth] = op; - pushBackObjects[pushBackDepth] = obj; - pushBackDepth++; + public final void pushBackToken(int op, Object obj) { + if (pushBackDepth >= pushBackInts.length - 1) { + int[] newInts = new int[pushBackInts.length * 2]; + System.arraycopy(pushBackInts, 0, newInts, 0, pushBackInts.length); + pushBackInts = newInts; + Object[] newObjects = new Object[pushBackObjects.length * 2]; + System.arraycopy(pushBackObjects, 0, newObjects, 0, pushBackObjects.length); + pushBackObjects = newObjects; + } + pushBackInts[pushBackDepth] = op; + pushBackObjects[pushBackDepth] = obj; + pushBackDepth++; } /** push back the most recently read token */ - public void pushBackToken() { pushBackToken(op, number != null ? (Object)number : (Object)string); } + public final void pushBackToken() { pushBackToken(op, number != null ? (Object)number : (Object)string); } /** read a token but leave it in the stream */ - public int peekToken() throws IOException { - int ret = getToken(); - pushBackToken(); - return ret; + public final int peekToken() throws IOException { + int ret = getToken(); + pushBackToken(); + return ret; } /** read a token */ - public int getToken() throws IOException { - number = null; - string = null; - if (pushBackDepth == 0) return op = _getToken(); - pushBackDepth--; - op = pushBackInts[pushBackDepth]; - if (pushBackObjects[pushBackDepth] != null) { - number = pushBackObjects[pushBackDepth] instanceof Number ? (Number)pushBackObjects[pushBackDepth] : null; - string = pushBackObjects[pushBackDepth] instanceof String ? (String)pushBackObjects[pushBackDepth] : null; - } - return op; + public final int getToken() throws IOException { + number = null; + string = null; + if (pushBackDepth == 0) { + mostRecentlyReadToken = op; + return op = _getToken(); + } + pushBackDepth--; + op = pushBackInts[pushBackDepth]; + if (pushBackObjects[pushBackDepth] != null) { + number = pushBackObjects[pushBackDepth] instanceof Number ? (Number)pushBackObjects[pushBackDepth] : null; + string = pushBackObjects[pushBackDepth] instanceof String ? (String)pushBackObjects[pushBackDepth] : null; + } + return op; } class LexerException extends IOException { - public LexerException(String s) { super(sourceName + ":" + line + "," + col + ": " + s); } + public LexerException(String s) { super(sourceName + ":" + line + "," + col + ": " + s); } } }